- Email: [email protected]
Ovarian carcinoma
Ovarian carcinoma: Improved survival following abdominopelvic irradiation in patients with a completed pelvic operation ALON
J. DEMBO,
RAYMOND FRANCIS A.
JOHN
M.D.
BEALE,
M.B.
A.
HELEN
M.B.
S. BUSH, BEAN,
M.D
F. PRINGLE,
JEREMY
M.B.
STURGEON,
JOAN
G.
REID,
Toronto,
Ontario,
M.D.
M.A.
Canada
A prospective, stratified, randomized study of 190 postoperative ovarian carcinoma patients with Stages 16, II, and Ill (asymptomatic) presentations is reported. The median time of follow-up was 52 months. Patients in whom bilateral salpingo-oophorectomy and hysterectomy (BSOH) could not be completed because of extensive pelvic tumor had a poor prognosis which did not differ for any of the therapies tested. When BSOH was completed, pelvic plus abdominopelvic irradiation (P+AB) with no diaphragmatic shielding significantly improved patient survival rate and long-term control of occult upper abdominal disease in approximately 25% more patients than pelvic irradiation alone or followed by adjuvant daily chlorambucil therapy. The effectiveness of P+AB in BSOH-ccmpleted patierits was independent of stage or tumor grade and was most clearly appreciated in patients with all gross tumor removed. Chlorambudl added to pelvic irradiation delayed the time to treatment failure, without reducing the number of treatment failures. (AM. J. OBSTET. GYNECOL. 134:793, 1979.)
E PIT HELIA L carcinoma of the ovary accounts for almost half the deaths from female genital tract malignancies in North America, with no trend of decreasing incidence or mortality rate being evident.’ Approximately two in three patients afflicted die of the disease.2 All three major therapeutic modalities, surgery, radiation therapy, and chemotherapy, have been reported to influence the natural course of the disease. However, the variable behavior of ovarian carcinoma has conFrom the Departments of Radiation, Medical Oncology, and ,Biostattitics, The Ontario Cancer Institute, incorporating The Princess Margaret Hospital. Presented at the Fourteenth Annual Meeting of the American Society of Clinical Oncology, Washington, D. C., April 3-4, 1978. Received Rewed Accepted
for publication August August
July
5, 1978.
22, 1978. 30,
1978.
Reprint requests: Dr. A. J. Dembo, The Princess Hos,bital, 500 Sherbourne St., Toronto, Ontario, M4.Y lK9. 0002-9378~7E~/150793+08$00.80/0~
1979 The
C. V.Mosby
Margaret Canada
Co.
founded most analyses of the precise benefit of each form of treatment, and it is not clear in which situations combined-modality therapy is superior to the use of one modality alone. ‘3 3 Powerful prognostic factors include the ability to complete bilateral salpingo-oophorectomy and hysterectomy (BSOH), the grade of the tumor, anatomic stage, the amount of residual tumor following surgery, and patient age and symptomatic status.2-4 In studies comparing therapies, if the patient groups are not closely matched with respect to these variables then a difference in results may be attributable to these factors rather than a difference in therapy. In 1971, Bush and co-workers,’ at The Princess Margaret Hospital in Toronto, began an ongoing study of recently diagnosed patients with invasive epithelial ovarian carcinoma of all stages. By December, 1975, 331 patients had been studied.” With the objective of better defining the optimum therapeutic strategy using existing techniques and modalities, the study design has been a stratified randomized one. The purpose of this article is to report on the results 793
794
Dembo et al.
No. Entered
UICC Stage
18
IB
132
II
40
Asympt. III
Treatment -.---..
t
lb 190
14106
Fig. 1. The design of the study and the number of patients in each stage and treatment randomization are shown. Fourteen U.I.C.C. Stage II patients would have been classified as International Federation of Gynecology and Obstetrics Stage III. of treatment of a subgroup of 190 patients IB, II, and asymptomatic III presentations.
with Stages
Material and methods The patient population. Between April, 197 1, and December, 1975, 199 patients with Stages IB, II, and III (asymptomatic) presentations were entered into the study. Asymptomatic is defined as the absence of abdominal or pelvic symptoms attributable to residual carcinoma or of ascites 4 to 6 weeks postoperatively.* Nine patients were excluded from this analysis (three protocol deviations, four refused treatment, and two had concurrent nonovarian carcinoma), leaving 190 patients as the population under study. The study was prospective and treatment was randomly allocated to patients stratified by age (above or below 50 years), Unio Internationalis Contra Cancrum (U.I.C.C.) stage classification, and pathologic type and grade (reviewed by Dr. T. C. Brown, Chief Pathologist, Princess Margaret Hospital, in all cases). Stratification by completeness of the initial pelvic operation (bilateral salpingo-oophorectomy and total abdominal hysterectomy. BSOH) was not undertaken before January. 1974, but has been since then, Completeness of operation refers only to BSOH and not to the presence or absence of residual tumor. Patients were excluded from the study: (1) if first seen more than 3 months postoperatively; (2) if postoperative systemic or radiation therapy had been begun elsewhere; (3) if so requested by patient or referring physician; or (4) if histopathology did not show invasive carcinoma (i.e., showed borderline malignancy). Patients were not excluded from the study because of advanced age, incomplete BSOH, gross residual disease, or bowel obstruction. Initial surgery was performed at a variety of hospitak prior to referral of patients to The Princess Margaret Hospital. Staging was based on the initial operative findings. Staging and *In retrospect, Stage III patients might better be divided as BSOH complete or incomplete, rather than asymptomatici symptomatic. Only two of 55 symptomatic Stage III patients seen during this period had BSOH completed. Survival of BSOH-incomplete but asymptomatic Stage III patients was similar to that of the symptomatic Stage III group.
restaging by peritoneoscopy or laparotomy were not performed on any patient. Study design and methods. The study design and the number of patients entered in each stage are shown in Fig. 1. When pelvic irradiation was given alone or with chlorambucil, the dose was 4,500 r-ads to the midplane in 20 fractions, using 15 by 15 cm opposing supervoltage beams. Abdominopelvic irradiation consisted of 2,250 rads to the midplane in 10 fractions given to the pelvis, followed immediately by 2.250 rads to the midplane in 10 fractions to the whole abdomen and pelvis, using cobalt with a caudad moving-strip technique, 24 to 30 cm wide by 10 cm long. The upper border of the abdominal held was at least 1 cm above the domes of the diaphragm, and no liver shielding was employed. The lower border was below the obturator foramen. The kidneys were shielded from the posterior beam throughout treatment. Chlorambucil, 6 mg per day (dosage modified according to blood counts), was given following pelvic irradiation for 2 years or until disease relapse to patients so randomized. Statistics. The end points of analysis were survival and relapse-free survival (based on nonoperative parameters). Both were dated from the date of diagnosis. Survival curves were calculated by actuarial methods and compared using the Ghan-modified Wilcoxon test. The frequency of various patterns of failure by treatment method were tested for the probability ot’ being similar by the Fisher exact test. A .5 ycat analysis is presented. Patients have been at risk 20 to X5 months, with a median of 52 month,.
Results An interim analysis suggested pelvic irradiation alone (PEL) to be inferior to the other two therapies, so patient entry into this limb of the study was discontinued after 43 patients had been accrued.’ In Fig. 2 the results of treatment are shown for the remaining 147 patients who received either pelvic irradiation followed by chlorambucil (P+CH) or pelvic plus abdominopelvic irradiation (P+AB). The trend in survival favoring P+AB over P+CH (60% versus 43%~ at 5 years) is not significant (p = 0. 13). I’he reason tar
Volume Number
Ovarian carcinoma
134 7
795
P’O.13
P <0.0005
IO' 0 Yrs.
after
diagnosis
, I
I 2
this will be shown to be the inclusion of the subgroup of patients with an incomplete BSOH. In Fig. 3 the survival characteristics for all three treatment methods are shown with the patients divided into two groups, based on the initial surgical procedure. The 5 year survival rate when BSOH could be completed was 65.7%, compared with 23.7% when BSOH was incomplete (usually due to extensive infiltrating pelvic tumor-tumor residuum was invariably larger than 2 cm in this group). A previous analysis showed this highly significant (p < 0.0005) difference in survival rates of over 35% to be independent of stage.4 It applies to Stage III (asymptomatic) and Stage IIB alike. Survival of the small proportion of patients with incomplete BSOH but no visible tumor residuum did not differ significantly from the survival of BSOHcompleted patients for any of the tested therapies. Thirty-six percent of the BSOH-completed group had visible tumor residuum, usually less than 2 cm diameter; 23% had no gross residuum. In 41% no tumor residuum was inferred to be present, although this was uncertain from the operative report. The clearest appreciation of the effect of postoperative treatment is obtained when BSOH-completed and BSOH-incomplete patient groups are considered separately. The poor prognosis for BSOH-incomplete patients is not appreciably different for any of the tested therapies. Four-year survival rate is below 30% for all three treatment methods. In contrast, the survival of BSOH-completed patients is markedly influenced by postoperative therapy and will be considered below in some detail. In Fig. 4 the relapse-free survival rate of BSOH-
I
I
4
5
Yrs. after diagnosis
14346
Fig. 2. Actuarial !survival curves for 147 patients, Stages IB, II, and III (asymptomatic), treated postoperatively with P+AB or P+CH. The numbers along the curves indicate the number of patients at risk at each time period; the error bars are 1 SD. The survival sca:leis logarithmic.
3
14316
Fig. 3. Actuarial survival curves for 190 patients, Stages IB, II, and III (asymptomatic), according to whether or not BSOH had been completed prior to randomization.
301 0
, I
2
1 3
4
Years after diagnosis
1 5
I
13719
Fig. 4. Actuarial relapse-free survival curves for 62 BSOHcompleted Stage II patients only, treated postoperatively with PEL or P+ AB.
completed Stage II patients only is compared for patients treated with pelvic irradiation (46%) versus pelvic plus abdominopelvic irradiation (79%). This difference is significant at the level p = 0.02, and will be explained below by the analysis of Table II. In Fig. 5 the survival curves for patients treated with P+AB and those treated with P+CH are shown for BSOH-completed, Stages IB, II, and III (asymptomatic) presentations only. At five years, the actuarial survival rate for P+ AB is 8 1% and only 5 1% for P+CH. These survival curves are different at a p significance level of 0.019. BSOH-completed patients comprise 69% of the total group (132 of 190 patients).
796
Dembo et al
Table I. Prognostic factors hv treatment inethod for 101 I~SOH-completed Stages IH, II, anti I II (as~tr11-‘to”‘at~c) patients (these fat-tors apply to the gr-oups shop in Fig. .i)
40
I I
1I
1 I I 2 3 4 Years after diagnosis
I
5 15696
Fig. 5. Actuarial survival curves for 101 BSOH-completed Stages IB, II, and III (asymptomatic) patients treated with P+AB or P+CH. In contrast to Fig. 2, when BSOHincomplete patients are excluded, the survival curves are significantly different.
4ge: Below 49 yr. Above 50 ~1’. Stage (LJ.I.C:.C:.): IB II 111 Histologic grade: Well-differentiated Poorly and moderately differentiated Histologic tvpc: Serous Mutinous Endometrioicl Other Residual tumor: Yes No
.33
54
26 3 16 ti
30 4 10 6
21 30 51
18 32 50
‘..n . . . . 9 . . . . A . . . . A . . . . aP+A8
(N-43-7)-
(~=31--6) Ye-ePEL a--oP+CH (N=42--8)
_
Table II. Analysis of treatment failures in 132 BSOH-completed Stages IB, II, and III (asymptomatic) patients.* Site of failure
-i
Pe1vi.s 2
abdomen
4oLy----J
5 Year uctuarial rela@e-fiee
Pelvb
clear
IYO.of 2 3 4 Years after diagnosis
I‘reatment
5 19706
Fig. 6. Actuarial survival for all three treatment methods loi 116 BSOH-completed Stages IB and II patients. Patients with Stage III were not treated with pelvic irradiation alone and hence are omitted. The numbers in parentheses indicate the numbers of patients at risk at times 0 and 5 years after diagnosis. The recognition of this patient classification is important, as it permits the identification of P-tAB as an improved postoperative treatment method which was not apparent in Fig. 2. From Table I it can be seen that the patients in each treatment group are closely matched for age, stage, tumor grade and type, and presence or absence of residual disease. These differences in survival can thus be reasonably attributed to the postoperative therapy given. In Fig. 6 actuarial survival is shown for all three treatment methods in Stages IB and II BSOH-completed patients. The apparent superiority of P+C;H over PEL suggested by the interim analysis was not
PELt P+CH P+AB
patients
31 51 50
No.
%
8 27 10 20 11 22
No.
%
,sunbal rate (%1
8 13f. ot
27 25 0
47 45 78
*Patients with pelvic disease (“pelvis +- abdomen”) who experienced treatment failure usually had upper abdominal disease as well. Those without pelvic disease at the time of relapse (“pelvis clear”) had upper abdominal-disease, ascites, or distant disease, with no demonstrable pelvic disease masses. tStages IB and I I only. sp < 0.0 I.
substantiated. For patients who had treatment failure, the median time to treatment failure was 13.6 months for PEL and 21.6 months for P+CH. However, this delay did not result in an improvement in survival rate at 5 years (58% for P+CH and 61% for PEL). These t.wo curves are not significantly different from each other. An improvement in survival should have been observed if the addition of chlorambucil produced permanent rather than temporary disease control. The survival of BSOH-completed patients treated by
Volume Number
Ovarian carcinoma
134 7
P+AB, appears, to be indeperident of stage. This is depicted in Fig. 7 where the survival of BSOH-completed Stage IB, Stage II, and Stage III’(asymptomatic) patients treated by P+AB is shown. Although patient numbers are small, no effect of stage on prognosis can be demonstrated. In Table II the sites of disease relapse are shown for BSOH-completed patients for all three treatment methods. These are based on nonoperative assessments i.e., clinical, radiologic, and sonographic evaluations. Patients are &ssified as having pelvic disease (with or without abdominal disease) at the time of relapse or as having no pelvic disease at the time of relapse. For all three treatment methods, the disease failure rate in the pelvis was similar, 20% to 27%. This is not surprising, since the dose of pelvic irradiation was the same for all patients. Houever, when there was no demonstrable pelvic tumor at the time of relapse, only patients receiving abdominal irradiation had long-term control of upper abdominal disease (masses and/or ascites). It is seen that the addition of abdominopelvic irradiation controls occult abdominal disease in approximately 25% more BSOH-completed patients than do the other two therapies. This significant (p < 0.01) reduction of disease failure in the abdomen is translated into an improvement in the relapse-free survival rate for P+AB treated patients, as shown in the last column of Table II. The impact in BSOH-completed patients of residual tumor on the frequency of tumor relapse is shown in Table III. P+AB patients with no (or uncertain) residuum had significantly fewer treatment failures than P+CH patients with no residuum or than patients with visible residual tumor receiving either therapy (p < 0.01). Another important observation is the interaction of therapy and the tumor grade, shown in Fig. 8. For P+CH patients with well-differentiated carcinomas the 5 year survival rate is 78%, compared with 43% for poorly and moderately differentiated tumors (p < 0.01). For P+AB patients the 5 year survival rate is 87% for well-differentiated and 77% for poorly differentiated tumors (p = 0.37). Thus, in contrast to chemotherapy the efficacy of P+AB is not inHuenced by tumor grade. For patients with poorly and moderately differentiated tumors, the 34% survival advantage for treatment P+AB compared with treatment P+CH is significant at a p level <0.02. Complications of treatment. Table IV lists in detail the complications of P+AB. Serious gastrointestinal toxicity was uncommon and did not occur in BSOHcompleted patients. Radiation may have contributed to the death of the one patient who required bowel surgery. She had multiple abdominal adhesions antedat-
501 0
1 I
I 2
n 3
797
I 5
I 4
Yrs. after diognosis
I4336
Fig. 7. Actuarial survival by stage for 50 BSOH-completed patients treated with P+AB only. Table III. Effect of tumor residuum frequency of tumor relapse in 101 BSOH-completed patients
on
.
No or uncertain residuum Tumor residuum present
3/32*
12130
8/18
II/21
*Lower than other three groups (p< 0.01). ing ovarian cancer. Autopsy showed recurrent carcinoma and mild radiation changes. In the PEL and P+CH treatment groups (114 patients) pelvic irradiation was associated with eight cases of gastrointestinal toxicity: three, occasional abdominal cramps without radiologic changes; two, rectal bleeding (proctitis); two required colostomy (one permanent); and one required resection of a bowel stricture plus a nephrectomy for urinary fistula, which probably contributed to death. Two of the three patients who had bowel surgery had received intrauterine irradiation (3,500 rads and 4,000 rads) in addition to their external irradiation. Overall, in two of 190 patients, death could be attributed to irradiation and both were BSOH incomplete. Thus GI toxicity from P+AB was no greater than from pelvic irradiation. Toxicities attributable to chlorambucil included seven cases of varicella zoster, three cases of major sepsis, four patients with nausea, and four cases of second neoplasm (one rectal carcinoma, one mixed parotid tumor, and two acute myeloid leukemia-leukemia developed 2 and 15 months, respectively, after cessation of chlorambucil therapy). BSOH-completed, P+CH-treated patients who ex-
798
Dembo
et al.
r
‘00 ._ go
\ --
7
P+AB
------.
Table IV. Complications P+AB irradiation
in 76 patients
rt&vlrlg
W.D.
(16-5) p-n.
-w -. .-----
80
too ‘t go-
‘.
‘.
80+ $ .? 70? a 602 $ 4" 50-
\\
W.D.
\\
(Ia- ,5) \\
'. . .\ \
\
p
\
\
\
\c
I ----.
1 P.D.
(33-4)
J
3Ot
20'
0
-I
1 I
I 2 Years
I 3 after
, 4 diagnosis
1 5
I IWO6
Fig. 8. Actuarial survival curves for the patients shown in Fig. 5, by treatment and tumor grade. (W. D.: Well-differentiated; I’. D.: poorly and moderately differentiated tumors.) P+AR treated patients with poorly differentiated tumors fared significantly better (p < 0.02) than did P+CH-treated patients with poorly differentiated tumors. perienced therapy “failures” received chlorambucil for a median of 72 weeks instead of 104 weeks as in the protocol. This shorter time is because the drug was discontinued at the time of disease relapse. Including episodes of temporary withdrawal, the median drug dosage was 75%, (4.5 mgiday) for this 72 week period. Nadir counts in 36% of patients were polymorphonuclear leukocytes 1,600 to 2,OOO/cu mm oi- platelets 101,000 to 15O,OOO/cu mm; in a further 45% polymorphonuclear leukocytes were less than 1,500icu mm or platelets less than 1OO,OOO/cu mm.
Comment The survival of patients in whom BSOH could not be completed because of the infiltrative nature of the pelvic tumor is poor. The prognosis of these patients is similar to those with advanced-stage presentations. Pelvic irradiation alone for such patients produces unsatisfactory results, which are not improved by the addition of abdominal irradiation or chlorambucil. (Jearlp, as for those with advanced disease, different therapeutic- strategies need to be sought for these pa-
Acute complications: Cramps and/or diarrhea Nausea and/or vomiting Cystitis* Myelosuppression Neutropenia < 3.0 X lo”/cu mm < 2.0 x 1ORICUn,m I‘hrombocytopenia < 150 X 19”icu mm < 100 X 103/cu mm < 50 x lo”/cu mm ‘I‘reatment dela) Due to myelosuppression (mean 28 days)+ Miscellaneous causes (mean 7 days): Chronic complications: Localized varicella zoster Radiologic basal lung fibrosis$ Asymptomatic raised alkaline phosphatase (> 12 KAU)! Clinical nonfatal radiation hepatitisll Radiation bowel damage Occasional diarrhea only Mild radiologic sigmoid colon stenosis Requiring surgery (died)# Single episcxie of hematuria Subsequent matignancy**
72
74 76
*Includes one infective case. tFour patients required reduction in radiation dose or volume. $‘I’wo nausea, two diarrhea, three absenteeism, one cardiac. $Symptomatic in one patient only (cough, pleurisy 2 months after irradiation). None required therapy: [here were no Fatalities. “Three of the patients had liver metastases at autopsy (KAU = King-Armstrong units). BTransient jaundice one year after irradiation. Pre-existing liver disease (thioridazine related?) #See text. **One ocular melanoma, one breast cancer, one diffuse histiocytic lymphoma of the thyroid, and one endometrial carcinoma (different histology). tients. incomplete BSOH due to tumor is a simple ob,jective means of identifying poor-prognosis patients, which is not predicted by standard staging classifications. Patients with ovarian carcinoma in whom BSOH can be completed account for approximately 55% of. postoperative patients of all stages referred to The Princess Margaret Hospital. The recognition of this subgroup of patients as having prognostically f’avorable disease rather than attempting to define prognosis by stage alone has been the major conceptual advanc:r gen-
Volume Number
134 7
erated by this study in planning therapeutic strategy. If with well-differentiated Stage IA those patients tumors, and those with Stage IV tumors are excluded, then over 45% of our postoperative patients could potentially benefit from treatment with pelvic plus abdominopelvic irradiation, with approximately 25% improvement in survival rate. The success of abdominopelvic irradiation in this study is probably related to the technique used. Great care is taken in localizing the superior border of the radiation field at least 1 cm above the domes of the diaphragm. Moreover, no liver shielding is employed. These two measures ensure that a major site of occult disease, the right undersurface of the diaphragm, is encompassed in the treatment plan.8 This approach necessitates a lower prescribed dose (2,250 rads in 10 fractions) tha.n others have used (i.e., 2,800 rads to 3,000 rads in 8 fractions93 ‘O), because of poor liver tolerance of high doses of irradiation. The data in Table II indicate that while not all occult upper abdominal disease is controlled by this dosage (many of the patients witF( disease failure in the pelvis also had upper abdominal disease) it is very uncommon to have disease failure in the upper abdomen when pelvic disease is controlled (0% in the present series). The relatively frequent biochemical evidence of liver damage and radiologic evidence of basal l&g damage (both rarely symptomatid) indicate that the technique we used does indeed encompass the entire abdominal cavity. Others using “whole abdominal” moving-strip irradiation have not reported these toxicities, suggesting that a major site of occult metastasis, the diaphragmatic undersurface, was not being treated.g* lo This, together with the frequent serious radiation bowel complications, and the excess of Stage IA patients receiving chemotherapy in the study of Smith and associates,” likely accounts for the different results they obtained. Their conclusion that chemotherapy is as effective as abdominal strip irradiation conflicts with our data (Fig. 5). The data in Fig. 4 indicate that pelvic irradiation alone is inadequate postoperative therapy in Stage II carcinoma of the ovary. W’hen a failure analysis similar to that in Table II is conducted for Stage II BSOHcompleted patients only, the following is found. With pelvic irradiation alone the frequency of disease failure in the abdomen beyond the pelvis is similar to the frequency of disease failure in the pelvis (6/27 versus S/27). The rationale for directing therapy beyond the pelvis is thus clear and was the basis for the present study.l Data in Fig. 4 and Table II show that the strategy of P+AB is effective treatment for unrecognized upper abdominal disease. The addition of low-dose, wide-
Ovarian carcinoma
799
fie!d total abdominal irradiation to pelvic irradiation decreases the frequency of disease failure beyond the pelvis, and this is reflected in the improved survival rate of P+AB-treated patients. The median time of follow-up of these patients (52 months) exceeds the median time of failure for patients who fail on pelvic irradiation alone (13 months) by a factor of 4. This suggests that the control of disease in the upper abdomen is likely to be permanent in most patients treated postoperatively with pelvic plus abdominopelvic irradiation. Thus a significant number of patients may be cured by the addition of abdominal irradiation to pelvic irradiation, if it is used as the standard postoperative therapy for Stage II patients. We believe this last statement is also applicable to Stage IB and Stage III patients with small amounts of tumor residuum. Using completeness of BSOH as the only criterion of “small residuum” it can be seen from Fig. 7 that stage is a very weak prognostic indicator under these circumstances. Herein lies the strongest argument against the prescription of therapy based on stage only. Results from failure analyses and peritoneoscopy reports would indicate that between 25% and 50% of the patients operatively staged as II might have been upstaged to Stage III if techniques of diaphragmatic visualization had been utilized.‘, 3. i As a consequence, there would likely have been more Stage III and fewer Stage II patients in Fig. 7, but the overall survival rate would remain unchanged, provided they were treated with P+AB. In Fig. 8 it is seen that postoperative pelvic plus abdominopelvic irradiation abolishes the negative prognostic effect of poorly (and moderately) differentiated histology in BSOH-completed’ patients. Such patients have a better survival expectation than those treated with pelvic irradiation followed by chlorambucil (77% versus 43%). The influence of histology is in accord with data from others which shows better survival rates with combination chemotherapy of well-differentiated ovarian neoplasms.” It is probably a reflection of the more favorable natural history of the well-differentiated tumors.’ Data from Table III showing the very favorable survival characteristics in P+AB patients with no \$sible residuum underscores the importance of surgical removal of all grossly visible disease where reasonably possible. Most of those with residual tumor had masses smaller than 2 cm. This indicates that for the reporting of treatment results three strata are required, i.e., no residuum (0), small residuum (~2 cm), and large residuum (>2 cm), rather than two strata (<2 cm/>2 cm), as is currently popular. The important negative finding in this study is that the policy of adjuvant chemotherapy with daily chlo-
800
Dembo et al.
rambucil following pelvic irradiation does itot improve long-term survi\~al. Chlorambucil added to pelvic iI-radiation delays the time to treatment failure b\ at1 average of 8 months, but does not reduce the numhct of’ treatment failures. ‘The hematologic toxicit) in the P+CH patients was mild to moderate. In retrospect ir might have been possible to adminstet, higher doses of drug. However, (1) there was no c.orrelation between drug dosage and f’requency of treatment tailure, (2) there N~S no correlation between 1ac.k of. hematologic roxicitv and treatment thilure, and (3) the 3 vca~ relapst-lice survival rate for the 30 Stages II and III no-residuum patients in this study was 34%. which is cerrainly no worse than a result of WA reported fi)r similar patients treated at the Mayo <:linic with cyclophosphamide with or without .4driamycin. ‘l‘hw while it is possible that chances ot’ long-tertn survival may have been improved with a mow aggressive polio. this cattnol simply be assumed to be so. and such 21 thesis would need to be verified in the setting of a controlled clinical trial. Similarly, while conclusions drawn front a policy of daily chlorambucil administration cannot be applied to intermittent melphalan therap), there arc* no data to show that response rates or survival is better with melphalan than \vith chlorambucil. A disquieting observation was the development of‘fatal acute myeloid leukemia in two ~hlorambu~il-tr-eatctl patients. Both were in the BSOH-completed group and were fi-ee of ovarian cancer. The significance of‘ this negative outcome in P+Clltreated patients must be considered in rhe context of the trend toward mow aggressive staging of‘ patien(s.H
‘l‘he cottclusion from staging perirotteoscrq)~ sttt(1ic.s t\ the same as from the f’ailurcs analysis in c~on~c~lttiott;tll\ sragcd patirttts (‘Table II), that i \. tnct;tst,tGc illtllOl III the uppet abdomen is ofien not w cigrikd 211 llw littw of‘ initial surgery. Howrvcr, to ;~ssunic tl1;11 Ilit. prosnosis and hcnw therap! of these paCent shwitl be rhc, samt as foi- paLients Xtaged diseast arc not the sitmc ;ik ,01iventionall> staged disease. I-he pt~cw~riprion oi ‘I aittgle alkylating agent in the l’ormer siluaticttt ma> dr-tly :I proportion of BSOH-completed pa’irnts the opportunit) of cut-c‘. For patients with recogrri~~cl or untwcgnized stnall amounts of‘ abdominal tlista~. 0u1 stud\ shows that pelvic plus abdominopel\,ic irradiatiott constitutes appropriate treatment with cnr;tC.c I)f~l~:rltiill (Fig.
7).
We acknowledge the assistance 01 the Biostatistit-s DCpartment of‘ The Ontario Cancer Ittstititte, the helpful criticisms of the clinical staf’f’(Drs. B. (:utrtrnings and 6. Thomas, in particular), and the willing cooperation in this study of the referring gynecologists.
REFERENCES
I. Siherberg, IL: Cancer statistics 197X, CA 28: 17, 1978. 2. Bagley, C. M., Jr., Young, R. C., Canellos, G. I’.. and 3. 4.
5.
6.
7.
DeVita, V. T.: Treatment of ovarian carcinoma: Possibiiities for progress, N. Engl. J. Med. 287~856, 1972. Tobias. 1. S.. and Griffiths. C. T.: Management of ovarian carcinoma. Current concepts and future prospects, N. Engi. J. Med. 294:818, 1976. Bush, R. S., Allt, W. E. C., Beale, F. A., Bean, H., Pringle, J. F., and Sturgeon, J,: Treatment of epithelial carcinoma of the ovary: Operation, irradiation, and chemotherapy. AM. J. OBSTET.GYNECOL. 127:692, 1977. Bush, R. S.: Carcinoma of ovary, in Bush, R. S., editor: Malignancies of the Ovary, Uterus and Cervix, England, Edward Arnold (Publishers) Ltd. In press. Ozols, R. F., Garvin, A. J., Costa, J., and Young, R. Cl.: <:orrelation of histologic grade with survival and response to therapy in epithelial ovarian cancer, Proc. Am. SOC. Clin. Oncol. 19:393, 1978. Mauch, P. M., Ehrmann, R. L., and Levine, M. B.: The significance of histologic grade in Stage II carcinoma of the ovary, Prtr. Am. Sot. Clin. Oncol. 19:312, 1978. ;II
”
Rosenoff, S. H.. DeVita, V. T., Hubbard, S., and Youn~g, R. C.: Peritoneoscopy in the staging and follou-up c;f ovarian cancer. Semin. Oflrol. 2:223. 1975. I,., and Smith, J. P.: Ovarian cancer, with special 9. Delclos, regard to types of radiotherapy, Nat]. Cancer inst. Monogr. 42: 129, 1975. IO. Perez, C. A., Korba, A., Zivnuska, I:.. et al.: “‘(:o Moving Strip Technique in the management of carcinoma of rhe ovary: Analysis of tumor control and morbidity, Int. ,J. Rad. Oncol. Biol. Phys. 4:379, 1978. Il. Smith, J. P.. Rutledge, F. N.. and Del&s. l..: Postoperative treatment of early cancer of rhe ovary: A random trial between postoperative irradiation and rhemotherspy, Natl. Cancer Inst. Monogr. 42: 149, 1975. 12. Edmonson, J. H., Fleming, T. R., Decker, D. (,.. CL al.: Different chemotherapeutic sensitivities and host factors affecting prognosis in advanced ovarian carcinoma versus minimal residual disease, Cancer ‘I’reat. Rep. 63:2,? 1, 1979. x.
J. DEMBO,
RAYMOND FRANCIS A.
JOHN
M.D.
BEALE,
M.B.
A.
HELEN
M.B.
S. BUSH, BEAN,
M.D
F. PRINGLE,
JEREMY
M.B.
STURGEON,
JOAN
G.
REID,
Toronto,
Ontario,
M.D.
M.A.
Canada
A prospective, stratified, randomized study of 190 postoperative ovarian carcinoma patients with Stages 16, II, and Ill (asymptomatic) presentations is reported. The median time of follow-up was 52 months. Patients in whom bilateral salpingo-oophorectomy and hysterectomy (BSOH) could not be completed because of extensive pelvic tumor had a poor prognosis which did not differ for any of the therapies tested. When BSOH was completed, pelvic plus abdominopelvic irradiation (P+AB) with no diaphragmatic shielding significantly improved patient survival rate and long-term control of occult upper abdominal disease in approximately 25% more patients than pelvic irradiation alone or followed by adjuvant daily chlorambucil therapy. The effectiveness of P+AB in BSOH-ccmpleted patierits was independent of stage or tumor grade and was most clearly appreciated in patients with all gross tumor removed. Chlorambudl added to pelvic irradiation delayed the time to treatment failure, without reducing the number of treatment failures. (AM. J. OBSTET. GYNECOL. 134:793, 1979.)
E PIT HELIA L carcinoma of the ovary accounts for almost half the deaths from female genital tract malignancies in North America, with no trend of decreasing incidence or mortality rate being evident.’ Approximately two in three patients afflicted die of the disease.2 All three major therapeutic modalities, surgery, radiation therapy, and chemotherapy, have been reported to influence the natural course of the disease. However, the variable behavior of ovarian carcinoma has conFrom the Departments of Radiation, Medical Oncology, and ,Biostattitics, The Ontario Cancer Institute, incorporating The Princess Margaret Hospital. Presented at the Fourteenth Annual Meeting of the American Society of Clinical Oncology, Washington, D. C., April 3-4, 1978. Received Rewed Accepted
for publication August August
July
5, 1978.
22, 1978. 30,
1978.
Reprint requests: Dr. A. J. Dembo, The Princess Hos,bital, 500 Sherbourne St., Toronto, Ontario, M4.Y lK9. 0002-9378~7E~/150793+08$00.80/0~
1979 The
C. V.Mosby
Margaret Canada
Co.
founded most analyses of the precise benefit of each form of treatment, and it is not clear in which situations combined-modality therapy is superior to the use of one modality alone. ‘3 3 Powerful prognostic factors include the ability to complete bilateral salpingo-oophorectomy and hysterectomy (BSOH), the grade of the tumor, anatomic stage, the amount of residual tumor following surgery, and patient age and symptomatic status.2-4 In studies comparing therapies, if the patient groups are not closely matched with respect to these variables then a difference in results may be attributable to these factors rather than a difference in therapy. In 1971, Bush and co-workers,’ at The Princess Margaret Hospital in Toronto, began an ongoing study of recently diagnosed patients with invasive epithelial ovarian carcinoma of all stages. By December, 1975, 331 patients had been studied.” With the objective of better defining the optimum therapeutic strategy using existing techniques and modalities, the study design has been a stratified randomized one. The purpose of this article is to report on the results 793
794
Dembo et al.
No. Entered
UICC Stage
18
IB
132
II
40
Asympt. III
Treatment -.---..
t
lb 190
14106
Fig. 1. The design of the study and the number of patients in each stage and treatment randomization are shown. Fourteen U.I.C.C. Stage II patients would have been classified as International Federation of Gynecology and Obstetrics Stage III. of treatment of a subgroup of 190 patients IB, II, and asymptomatic III presentations.
with Stages
Material and methods The patient population. Between April, 197 1, and December, 1975, 199 patients with Stages IB, II, and III (asymptomatic) presentations were entered into the study. Asymptomatic is defined as the absence of abdominal or pelvic symptoms attributable to residual carcinoma or of ascites 4 to 6 weeks postoperatively.* Nine patients were excluded from this analysis (three protocol deviations, four refused treatment, and two had concurrent nonovarian carcinoma), leaving 190 patients as the population under study. The study was prospective and treatment was randomly allocated to patients stratified by age (above or below 50 years), Unio Internationalis Contra Cancrum (U.I.C.C.) stage classification, and pathologic type and grade (reviewed by Dr. T. C. Brown, Chief Pathologist, Princess Margaret Hospital, in all cases). Stratification by completeness of the initial pelvic operation (bilateral salpingo-oophorectomy and total abdominal hysterectomy. BSOH) was not undertaken before January. 1974, but has been since then, Completeness of operation refers only to BSOH and not to the presence or absence of residual tumor. Patients were excluded from the study: (1) if first seen more than 3 months postoperatively; (2) if postoperative systemic or radiation therapy had been begun elsewhere; (3) if so requested by patient or referring physician; or (4) if histopathology did not show invasive carcinoma (i.e., showed borderline malignancy). Patients were not excluded from the study because of advanced age, incomplete BSOH, gross residual disease, or bowel obstruction. Initial surgery was performed at a variety of hospitak prior to referral of patients to The Princess Margaret Hospital. Staging was based on the initial operative findings. Staging and *In retrospect, Stage III patients might better be divided as BSOH complete or incomplete, rather than asymptomatici symptomatic. Only two of 55 symptomatic Stage III patients seen during this period had BSOH completed. Survival of BSOH-incomplete but asymptomatic Stage III patients was similar to that of the symptomatic Stage III group.
restaging by peritoneoscopy or laparotomy were not performed on any patient. Study design and methods. The study design and the number of patients entered in each stage are shown in Fig. 1. When pelvic irradiation was given alone or with chlorambucil, the dose was 4,500 r-ads to the midplane in 20 fractions, using 15 by 15 cm opposing supervoltage beams. Abdominopelvic irradiation consisted of 2,250 rads to the midplane in 10 fractions given to the pelvis, followed immediately by 2.250 rads to the midplane in 10 fractions to the whole abdomen and pelvis, using cobalt with a caudad moving-strip technique, 24 to 30 cm wide by 10 cm long. The upper border of the abdominal held was at least 1 cm above the domes of the diaphragm, and no liver shielding was employed. The lower border was below the obturator foramen. The kidneys were shielded from the posterior beam throughout treatment. Chlorambucil, 6 mg per day (dosage modified according to blood counts), was given following pelvic irradiation for 2 years or until disease relapse to patients so randomized. Statistics. The end points of analysis were survival and relapse-free survival (based on nonoperative parameters). Both were dated from the date of diagnosis. Survival curves were calculated by actuarial methods and compared using the Ghan-modified Wilcoxon test. The frequency of various patterns of failure by treatment method were tested for the probability ot’ being similar by the Fisher exact test. A .5 ycat analysis is presented. Patients have been at risk 20 to X5 months, with a median of 52 month,.
Results An interim analysis suggested pelvic irradiation alone (PEL) to be inferior to the other two therapies, so patient entry into this limb of the study was discontinued after 43 patients had been accrued.’ In Fig. 2 the results of treatment are shown for the remaining 147 patients who received either pelvic irradiation followed by chlorambucil (P+CH) or pelvic plus abdominopelvic irradiation (P+AB). The trend in survival favoring P+AB over P+CH (60% versus 43%~ at 5 years) is not significant (p = 0. 13). I’he reason tar
Volume Number
Ovarian carcinoma
134 7
795
P’O.13
P <0.0005
IO' 0 Yrs.
after
diagnosis
, I
I 2
this will be shown to be the inclusion of the subgroup of patients with an incomplete BSOH. In Fig. 3 the survival characteristics for all three treatment methods are shown with the patients divided into two groups, based on the initial surgical procedure. The 5 year survival rate when BSOH could be completed was 65.7%, compared with 23.7% when BSOH was incomplete (usually due to extensive infiltrating pelvic tumor-tumor residuum was invariably larger than 2 cm in this group). A previous analysis showed this highly significant (p < 0.0005) difference in survival rates of over 35% to be independent of stage.4 It applies to Stage III (asymptomatic) and Stage IIB alike. Survival of the small proportion of patients with incomplete BSOH but no visible tumor residuum did not differ significantly from the survival of BSOHcompleted patients for any of the tested therapies. Thirty-six percent of the BSOH-completed group had visible tumor residuum, usually less than 2 cm diameter; 23% had no gross residuum. In 41% no tumor residuum was inferred to be present, although this was uncertain from the operative report. The clearest appreciation of the effect of postoperative treatment is obtained when BSOH-completed and BSOH-incomplete patient groups are considered separately. The poor prognosis for BSOH-incomplete patients is not appreciably different for any of the tested therapies. Four-year survival rate is below 30% for all three treatment methods. In contrast, the survival of BSOH-completed patients is markedly influenced by postoperative therapy and will be considered below in some detail. In Fig. 4 the relapse-free survival rate of BSOH-
I
I
4
5
Yrs. after diagnosis
14346
Fig. 2. Actuarial !survival curves for 147 patients, Stages IB, II, and III (asymptomatic), treated postoperatively with P+AB or P+CH. The numbers along the curves indicate the number of patients at risk at each time period; the error bars are 1 SD. The survival sca:leis logarithmic.
3
14316
Fig. 3. Actuarial survival curves for 190 patients, Stages IB, II, and III (asymptomatic), according to whether or not BSOH had been completed prior to randomization.
301 0
, I
2
1 3
4
Years after diagnosis
1 5
I
13719
Fig. 4. Actuarial relapse-free survival curves for 62 BSOHcompleted Stage II patients only, treated postoperatively with PEL or P+ AB.
completed Stage II patients only is compared for patients treated with pelvic irradiation (46%) versus pelvic plus abdominopelvic irradiation (79%). This difference is significant at the level p = 0.02, and will be explained below by the analysis of Table II. In Fig. 5 the survival curves for patients treated with P+AB and those treated with P+CH are shown for BSOH-completed, Stages IB, II, and III (asymptomatic) presentations only. At five years, the actuarial survival rate for P+ AB is 8 1% and only 5 1% for P+CH. These survival curves are different at a p significance level of 0.019. BSOH-completed patients comprise 69% of the total group (132 of 190 patients).
796
Dembo et al
Table I. Prognostic factors hv treatment inethod for 101 I~SOH-completed Stages IH, II, anti I II (as~tr11-‘to”‘at~c) patients (these fat-tors apply to the gr-oups shop in Fig. .i)
40
I I
1I
1 I I 2 3 4 Years after diagnosis
I
5 15696
Fig. 5. Actuarial survival curves for 101 BSOH-completed Stages IB, II, and III (asymptomatic) patients treated with P+AB or P+CH. In contrast to Fig. 2, when BSOHincomplete patients are excluded, the survival curves are significantly different.
4ge: Below 49 yr. Above 50 ~1’. Stage (LJ.I.C:.C:.): IB II 111 Histologic grade: Well-differentiated Poorly and moderately differentiated Histologic tvpc: Serous Mutinous Endometrioicl Other Residual tumor: Yes No
.33
54
26 3 16 ti
30 4 10 6
21 30 51
18 32 50
‘..n . . . . 9 . . . . A . . . . A . . . . aP+A8
(N-43-7)-
(~=31--6) Ye-ePEL a--oP+CH (N=42--8)
_
Table II. Analysis of treatment failures in 132 BSOH-completed Stages IB, II, and III (asymptomatic) patients.* Site of failure
-i
Pe1vi.s 2
abdomen
4oLy----J
5 Year uctuarial rela@e-fiee
Pelvb
clear
IYO.of 2 3 4 Years after diagnosis
I‘reatment
5 19706
Fig. 6. Actuarial survival for all three treatment methods loi 116 BSOH-completed Stages IB and II patients. Patients with Stage III were not treated with pelvic irradiation alone and hence are omitted. The numbers in parentheses indicate the numbers of patients at risk at times 0 and 5 years after diagnosis. The recognition of this patient classification is important, as it permits the identification of P-tAB as an improved postoperative treatment method which was not apparent in Fig. 2. From Table I it can be seen that the patients in each treatment group are closely matched for age, stage, tumor grade and type, and presence or absence of residual disease. These differences in survival can thus be reasonably attributed to the postoperative therapy given. In Fig. 6 actuarial survival is shown for all three treatment methods in Stages IB and II BSOH-completed patients. The apparent superiority of P+C;H over PEL suggested by the interim analysis was not
PELt P+CH P+AB
patients
31 51 50
No.
%
8 27 10 20 11 22
No.
%
,sunbal rate (%1
8 13f. ot
27 25 0
47 45 78
*Patients with pelvic disease (“pelvis +- abdomen”) who experienced treatment failure usually had upper abdominal disease as well. Those without pelvic disease at the time of relapse (“pelvis clear”) had upper abdominal-disease, ascites, or distant disease, with no demonstrable pelvic disease masses. tStages IB and I I only. sp < 0.0 I.
substantiated. For patients who had treatment failure, the median time to treatment failure was 13.6 months for PEL and 21.6 months for P+CH. However, this delay did not result in an improvement in survival rate at 5 years (58% for P+CH and 61% for PEL). These t.wo curves are not significantly different from each other. An improvement in survival should have been observed if the addition of chlorambucil produced permanent rather than temporary disease control. The survival of BSOH-completed patients treated by
Volume Number
Ovarian carcinoma
134 7
P+AB, appears, to be indeperident of stage. This is depicted in Fig. 7 where the survival of BSOH-completed Stage IB, Stage II, and Stage III’(asymptomatic) patients treated by P+AB is shown. Although patient numbers are small, no effect of stage on prognosis can be demonstrated. In Table II the sites of disease relapse are shown for BSOH-completed patients for all three treatment methods. These are based on nonoperative assessments i.e., clinical, radiologic, and sonographic evaluations. Patients are &ssified as having pelvic disease (with or without abdominal disease) at the time of relapse or as having no pelvic disease at the time of relapse. For all three treatment methods, the disease failure rate in the pelvis was similar, 20% to 27%. This is not surprising, since the dose of pelvic irradiation was the same for all patients. Houever, when there was no demonstrable pelvic tumor at the time of relapse, only patients receiving abdominal irradiation had long-term control of upper abdominal disease (masses and/or ascites). It is seen that the addition of abdominopelvic irradiation controls occult abdominal disease in approximately 25% more BSOH-completed patients than do the other two therapies. This significant (p < 0.01) reduction of disease failure in the abdomen is translated into an improvement in the relapse-free survival rate for P+AB treated patients, as shown in the last column of Table II. The impact in BSOH-completed patients of residual tumor on the frequency of tumor relapse is shown in Table III. P+AB patients with no (or uncertain) residuum had significantly fewer treatment failures than P+CH patients with no residuum or than patients with visible residual tumor receiving either therapy (p < 0.01). Another important observation is the interaction of therapy and the tumor grade, shown in Fig. 8. For P+CH patients with well-differentiated carcinomas the 5 year survival rate is 78%, compared with 43% for poorly and moderately differentiated tumors (p < 0.01). For P+AB patients the 5 year survival rate is 87% for well-differentiated and 77% for poorly differentiated tumors (p = 0.37). Thus, in contrast to chemotherapy the efficacy of P+AB is not inHuenced by tumor grade. For patients with poorly and moderately differentiated tumors, the 34% survival advantage for treatment P+AB compared with treatment P+CH is significant at a p level <0.02. Complications of treatment. Table IV lists in detail the complications of P+AB. Serious gastrointestinal toxicity was uncommon and did not occur in BSOHcompleted patients. Radiation may have contributed to the death of the one patient who required bowel surgery. She had multiple abdominal adhesions antedat-
501 0
1 I
I 2
n 3
797
I 5
I 4
Yrs. after diognosis
I4336
Fig. 7. Actuarial survival by stage for 50 BSOH-completed patients treated with P+AB only. Table III. Effect of tumor residuum frequency of tumor relapse in 101 BSOH-completed patients
on
.
No or uncertain residuum Tumor residuum present
3/32*
12130
8/18
II/21
*Lower than other three groups (p< 0.01). ing ovarian cancer. Autopsy showed recurrent carcinoma and mild radiation changes. In the PEL and P+CH treatment groups (114 patients) pelvic irradiation was associated with eight cases of gastrointestinal toxicity: three, occasional abdominal cramps without radiologic changes; two, rectal bleeding (proctitis); two required colostomy (one permanent); and one required resection of a bowel stricture plus a nephrectomy for urinary fistula, which probably contributed to death. Two of the three patients who had bowel surgery had received intrauterine irradiation (3,500 rads and 4,000 rads) in addition to their external irradiation. Overall, in two of 190 patients, death could be attributed to irradiation and both were BSOH incomplete. Thus GI toxicity from P+AB was no greater than from pelvic irradiation. Toxicities attributable to chlorambucil included seven cases of varicella zoster, three cases of major sepsis, four patients with nausea, and four cases of second neoplasm (one rectal carcinoma, one mixed parotid tumor, and two acute myeloid leukemia-leukemia developed 2 and 15 months, respectively, after cessation of chlorambucil therapy). BSOH-completed, P+CH-treated patients who ex-
798
Dembo
et al.
r
‘00 ._ go
\ --
7
P+AB
------.
Table IV. Complications P+AB irradiation
in 76 patients
rt&vlrlg
W.D.
(16-5) p-n.
-w -. .-----
80
too ‘t go-
‘.
‘.
80+ $ .? 70? a 602 $ 4" 50-
\\
W.D.
\\
(Ia- ,5) \\
'. . .\ \
\
p
\
\
\
\c
I ----.
1 P.D.
(33-4)
J
3Ot
20'
0
-I
1 I
I 2 Years
I 3 after
, 4 diagnosis
1 5
I IWO6
Fig. 8. Actuarial survival curves for the patients shown in Fig. 5, by treatment and tumor grade. (W. D.: Well-differentiated; I’. D.: poorly and moderately differentiated tumors.) P+AR treated patients with poorly differentiated tumors fared significantly better (p < 0.02) than did P+CH-treated patients with poorly differentiated tumors. perienced therapy “failures” received chlorambucil for a median of 72 weeks instead of 104 weeks as in the protocol. This shorter time is because the drug was discontinued at the time of disease relapse. Including episodes of temporary withdrawal, the median drug dosage was 75%, (4.5 mgiday) for this 72 week period. Nadir counts in 36% of patients were polymorphonuclear leukocytes 1,600 to 2,OOO/cu mm oi- platelets 101,000 to 15O,OOO/cu mm; in a further 45% polymorphonuclear leukocytes were less than 1,500icu mm or platelets less than 1OO,OOO/cu mm.
Comment The survival of patients in whom BSOH could not be completed because of the infiltrative nature of the pelvic tumor is poor. The prognosis of these patients is similar to those with advanced-stage presentations. Pelvic irradiation alone for such patients produces unsatisfactory results, which are not improved by the addition of abdominal irradiation or chlorambucil. (Jearlp, as for those with advanced disease, different therapeutic- strategies need to be sought for these pa-
Acute complications: Cramps and/or diarrhea Nausea and/or vomiting Cystitis* Myelosuppression Neutropenia < 3.0 X lo”/cu mm < 2.0 x 1ORICUn,m I‘hrombocytopenia < 150 X 19”icu mm < 100 X 103/cu mm < 50 x lo”/cu mm ‘I‘reatment dela) Due to myelosuppression (mean 28 days)+ Miscellaneous causes (mean 7 days): Chronic complications: Localized varicella zoster Radiologic basal lung fibrosis$ Asymptomatic raised alkaline phosphatase (> 12 KAU)! Clinical nonfatal radiation hepatitisll Radiation bowel damage Occasional diarrhea only Mild radiologic sigmoid colon stenosis Requiring surgery (died)# Single episcxie of hematuria Subsequent matignancy**
72
74 76
*Includes one infective case. tFour patients required reduction in radiation dose or volume. $‘I’wo nausea, two diarrhea, three absenteeism, one cardiac. $Symptomatic in one patient only (cough, pleurisy 2 months after irradiation). None required therapy: [here were no Fatalities. “Three of the patients had liver metastases at autopsy (KAU = King-Armstrong units). BTransient jaundice one year after irradiation. Pre-existing liver disease (thioridazine related?) #See text. **One ocular melanoma, one breast cancer, one diffuse histiocytic lymphoma of the thyroid, and one endometrial carcinoma (different histology). tients. incomplete BSOH due to tumor is a simple ob,jective means of identifying poor-prognosis patients, which is not predicted by standard staging classifications. Patients with ovarian carcinoma in whom BSOH can be completed account for approximately 55% of. postoperative patients of all stages referred to The Princess Margaret Hospital. The recognition of this subgroup of patients as having prognostically f’avorable disease rather than attempting to define prognosis by stage alone has been the major conceptual advanc:r gen-
Volume Number
134 7
erated by this study in planning therapeutic strategy. If with well-differentiated Stage IA those patients tumors, and those with Stage IV tumors are excluded, then over 45% of our postoperative patients could potentially benefit from treatment with pelvic plus abdominopelvic irradiation, with approximately 25% improvement in survival rate. The success of abdominopelvic irradiation in this study is probably related to the technique used. Great care is taken in localizing the superior border of the radiation field at least 1 cm above the domes of the diaphragm. Moreover, no liver shielding is employed. These two measures ensure that a major site of occult disease, the right undersurface of the diaphragm, is encompassed in the treatment plan.8 This approach necessitates a lower prescribed dose (2,250 rads in 10 fractions) tha.n others have used (i.e., 2,800 rads to 3,000 rads in 8 fractions93 ‘O), because of poor liver tolerance of high doses of irradiation. The data in Table II indicate that while not all occult upper abdominal disease is controlled by this dosage (many of the patients witF( disease failure in the pelvis also had upper abdominal disease) it is very uncommon to have disease failure in the upper abdomen when pelvic disease is controlled (0% in the present series). The relatively frequent biochemical evidence of liver damage and radiologic evidence of basal l&g damage (both rarely symptomatid) indicate that the technique we used does indeed encompass the entire abdominal cavity. Others using “whole abdominal” moving-strip irradiation have not reported these toxicities, suggesting that a major site of occult metastasis, the diaphragmatic undersurface, was not being treated.g* lo This, together with the frequent serious radiation bowel complications, and the excess of Stage IA patients receiving chemotherapy in the study of Smith and associates,” likely accounts for the different results they obtained. Their conclusion that chemotherapy is as effective as abdominal strip irradiation conflicts with our data (Fig. 5). The data in Fig. 4 indicate that pelvic irradiation alone is inadequate postoperative therapy in Stage II carcinoma of the ovary. W’hen a failure analysis similar to that in Table II is conducted for Stage II BSOHcompleted patients only, the following is found. With pelvic irradiation alone the frequency of disease failure in the abdomen beyond the pelvis is similar to the frequency of disease failure in the pelvis (6/27 versus S/27). The rationale for directing therapy beyond the pelvis is thus clear and was the basis for the present study.l Data in Fig. 4 and Table II show that the strategy of P+AB is effective treatment for unrecognized upper abdominal disease. The addition of low-dose, wide-
Ovarian carcinoma
799
fie!d total abdominal irradiation to pelvic irradiation decreases the frequency of disease failure beyond the pelvis, and this is reflected in the improved survival rate of P+AB-treated patients. The median time of follow-up of these patients (52 months) exceeds the median time of failure for patients who fail on pelvic irradiation alone (13 months) by a factor of 4. This suggests that the control of disease in the upper abdomen is likely to be permanent in most patients treated postoperatively with pelvic plus abdominopelvic irradiation. Thus a significant number of patients may be cured by the addition of abdominal irradiation to pelvic irradiation, if it is used as the standard postoperative therapy for Stage II patients. We believe this last statement is also applicable to Stage IB and Stage III patients with small amounts of tumor residuum. Using completeness of BSOH as the only criterion of “small residuum” it can be seen from Fig. 7 that stage is a very weak prognostic indicator under these circumstances. Herein lies the strongest argument against the prescription of therapy based on stage only. Results from failure analyses and peritoneoscopy reports would indicate that between 25% and 50% of the patients operatively staged as II might have been upstaged to Stage III if techniques of diaphragmatic visualization had been utilized.‘, 3. i As a consequence, there would likely have been more Stage III and fewer Stage II patients in Fig. 7, but the overall survival rate would remain unchanged, provided they were treated with P+AB. In Fig. 8 it is seen that postoperative pelvic plus abdominopelvic irradiation abolishes the negative prognostic effect of poorly (and moderately) differentiated histology in BSOH-completed’ patients. Such patients have a better survival expectation than those treated with pelvic irradiation followed by chlorambucil (77% versus 43%). The influence of histology is in accord with data from others which shows better survival rates with combination chemotherapy of well-differentiated ovarian neoplasms.” It is probably a reflection of the more favorable natural history of the well-differentiated tumors.’ Data from Table III showing the very favorable survival characteristics in P+AB patients with no \$sible residuum underscores the importance of surgical removal of all grossly visible disease where reasonably possible. Most of those with residual tumor had masses smaller than 2 cm. This indicates that for the reporting of treatment results three strata are required, i.e., no residuum (0), small residuum (~2 cm), and large residuum (>2 cm), rather than two strata (<2 cm/>2 cm), as is currently popular. The important negative finding in this study is that the policy of adjuvant chemotherapy with daily chlo-
800
Dembo et al.
rambucil following pelvic irradiation does itot improve long-term survi\~al. Chlorambucil added to pelvic iI-radiation delays the time to treatment failure b\ at1 average of 8 months, but does not reduce the numhct of’ treatment failures. ‘The hematologic toxicit) in the P+CH patients was mild to moderate. In retrospect ir might have been possible to adminstet, higher doses of drug. However, (1) there was no c.orrelation between drug dosage and f’requency of treatment tailure, (2) there N~S no correlation between 1ac.k of. hematologic roxicitv and treatment thilure, and (3) the 3 vca~ relapst-lice survival rate for the 30 Stages II and III no-residuum patients in this study was 34%. which is cerrainly no worse than a result of WA reported fi)r similar patients treated at the Mayo <:linic with cyclophosphamide with or without .4driamycin. ‘l‘hw while it is possible that chances ot’ long-tertn survival may have been improved with a mow aggressive polio. this cattnol simply be assumed to be so. and such 21 thesis would need to be verified in the setting of a controlled clinical trial. Similarly, while conclusions drawn front a policy of daily chlorambucil administration cannot be applied to intermittent melphalan therap), there arc* no data to show that response rates or survival is better with melphalan than \vith chlorambucil. A disquieting observation was the development of‘fatal acute myeloid leukemia in two ~hlorambu~il-tr-eatctl patients. Both were in the BSOH-completed group and were fi-ee of ovarian cancer. The significance of‘ this negative outcome in P+Clltreated patients must be considered in rhe context of the trend toward mow aggressive staging of‘ patien(s.H
‘l‘he cottclusion from staging perirotteoscrq)~ sttt(1ic.s t\ the same as from the f’ailurcs analysis in c~on~c~lttiott;tll\ sragcd patirttts (‘Table II), that i \. tnct;tst,tGc illtllOl III the uppet abdomen is ofien not w cigrikd 211 llw littw of‘ initial surgery. Howrvcr, to ;~ssunic tl1;11 Ilit. prosnosis and hcnw therap! of these paCent shwitl be rhc, samt as foi- paLients X
7).
We acknowledge the assistance 01 the Biostatistit-s DCpartment of‘ The Ontario Cancer Ittstititte, the helpful criticisms of the clinical staf’f’(Drs. B. (:utrtrnings and 6. Thomas, in particular), and the willing cooperation in this study of the referring gynecologists.
REFERENCES
I. Siherberg, IL: Cancer statistics 197X, CA 28: 17, 1978. 2. Bagley, C. M., Jr., Young, R. C., Canellos, G. I’.. and 3. 4.
5.
6.
7.
DeVita, V. T.: Treatment of ovarian carcinoma: Possibiiities for progress, N. Engl. J. Med. 287~856, 1972. Tobias. 1. S.. and Griffiths. C. T.: Management of ovarian carcinoma. Current concepts and future prospects, N. Engi. J. Med. 294:818, 1976. Bush, R. S., Allt, W. E. C., Beale, F. A., Bean, H., Pringle, J. F., and Sturgeon, J,: Treatment of epithelial carcinoma of the ovary: Operation, irradiation, and chemotherapy. AM. J. OBSTET.GYNECOL. 127:692, 1977. Bush, R. S.: Carcinoma of ovary, in Bush, R. S., editor: Malignancies of the Ovary, Uterus and Cervix, England, Edward Arnold (Publishers) Ltd. In press. Ozols, R. F., Garvin, A. J., Costa, J., and Young, R. Cl.: <:orrelation of histologic grade with survival and response to therapy in epithelial ovarian cancer, Proc. Am. SOC. Clin. Oncol. 19:393, 1978. Mauch, P. M., Ehrmann, R. L., and Levine, M. B.: The significance of histologic grade in Stage II carcinoma of the ovary, Prtr. Am. Sot. Clin. Oncol. 19:312, 1978. ;II
”
Rosenoff, S. H.. DeVita, V. T., Hubbard, S., and Youn~g, R. C.: Peritoneoscopy in the staging and follou-up c;f ovarian cancer. Semin. Oflrol. 2:223. 1975. I,., and Smith, J. P.: Ovarian cancer, with special 9. Delclos, regard to types of radiotherapy, Nat]. Cancer inst. Monogr. 42: 129, 1975. IO. Perez, C. A., Korba, A., Zivnuska, I:.. et al.: “‘(:o Moving Strip Technique in the management of carcinoma of rhe ovary: Analysis of tumor control and morbidity, Int. ,J. Rad. Oncol. Biol. Phys. 4:379, 1978. Il. Smith, J. P.. Rutledge, F. N.. and Del&s. l..: Postoperative treatment of early cancer of rhe ovary: A random trial between postoperative irradiation and rhemotherspy, Natl. Cancer Inst. Monogr. 42: 149, 1975. 12. Edmonson, J. H., Fleming, T. R., Decker, D. (,.. CL al.: Different chemotherapeutic sensitivities and host factors affecting prognosis in advanced ovarian carcinoma versus minimal residual disease, Cancer ‘I’reat. Rep. 63:2,? 1, 1979. x.